Your vote, squared: not everyone's vote is equal - insight into quadratic voting and why there wasn't linear rewards in the first place

in #steem4 years ago (edited)

Research by Eric Posner and Glen Weyl into the realm of quadratic voting has shown interesting promise in corporate shareholder votes. Does quadratic voting (pre-HF19) have a place here on steem?

With hardfork 19 right around the corner, I wanted to do a short post detailing quadratic voting and it's place here on steem and in voting in general.

Quadratic Voting is a concept developed by Glen Weyl and Eric Posner, both professors at the University of Chicago. They wrote "Quadratic Vote Buying as Efficient Corporate Governance" detailing what has been termed "the most important idea of law and public policy that has emerged from economics in the last ten years". Weyl and Eric were immediately fascinated by this idea's application to corporate governance; shareholders govern by buying votes for a price that is the square of the number of votes.

Although no voting process is perfectly efficient, there are problems with the simple one-share, one-vote, majority-wins mentality. The "tyranny of the majority" often results when a large number of people only care very little about an outcome overpower a minority who cares passionately.

When you make people pay for the square of their vote, this causes actors with little preference to maybe buy one or two votes either way. If you are very passionate about an outcome, you are able to buy enough shares to influence that decision - although at the cost of pivotally influencing the vote against the will of many others. The increased cost of making such votes in quadratic voting ensures that voters only do this when they are very sure they are right.

Eric Posner writes in his 2014 blog post about the real value of this voting style in the terms of democratic society.

But the most important setting is democracy itself. An incredibly complicated system of institutional self-checking (separation of powers, federalism) and judicially enforced constitutional rights try to correct for the defects of one-person-one-vote, but do so very badly. Can quadratic voting do better? Glen and I argue that it can.

You can read more about quadratic voting from this interview and read Eric and Glen's paper here

One of the problems in steem is even though the post payouts were quadratic, giving those with lots of steem power more influence, the cost to the whales is not reflected in steem's economic model. The quadratic reward structure is explained in the steem white paper below:

In order to realign incentives and discourage individuals from simply voting for themselves, money must be distributed in a nonlinear manner. [...] Assuming all users have equal stake, someone who only receives their own vote will receive much less than someone who receives votes from 100 different users. This encourages users to cooperate to vote for the same things to maximize the payout. This system also creates financial incentive to collude where everyone votes on one thing and then divides the reward equally among themselves.

It is not like voting with a high weight uses exponentially more of your vote power.

So to be honest, I was never sure what to make of this section of the white paper. I understand the motivations but not necessarily the implementation. Some of the clear problems are that users do not have equal stake and that the system is operating far from equilibrium. The incentives to cooperate around certain posts is painfully obvious however. This is seemingly designed almost entirely to prevent the issues with everyone self-voting; an issue that might not even be proper to address.

Given the complexity of steem's economics, it will really be interesting to see how the community reacts to the changes in HF19. I am mostly in favor of both linear voting and the extra power; I hope to see more quality content rewarded properly. We are learning a lot as a community how these untested economics work at scale in the real world. - this is not the final implementation. Remember MySpace.

Stay decentralized,


Helpful thoughts, as usual. I was unaware about the U of Chicago research in this area. Part of the challenge in any type of analysis in these problems is what you mentioned, they're just so complex with so many different potential motives. For example, with or without the HF19 , I'm personally unlikely to change my voting behavior. Rather than vote for myself with all my upvotes, which is something I typically avoid unless I want to improve visibility of my comment for example, I like to upvote other people and TRY to help build a sense of community. Largely this action is somewhat selfish on my part, because I would like to see Steemit grow and see my stake become a lot more valuable.

Please execute the people that implemented exponential influence :)

How the fuck could anyone have the balls to give someone with 10 times more money, 100 times more influence, REALLY? It should literally be the other way around, someone with 100x more money should only have 10 times more influence. And even that is insanely high in it's current implementation because curation quality is completely ignored. A person with 10 times more SP should also be 10 times more critical and should be penalized 10 times as hard for making curation mistakes.


Did you read my blog? The linked academic papers?

The point is exactly that, to give those who care deeply (and thus willing to spend more) more influence than those who do not care as much. The extra money spent by the people who care more is redistributed to all shareholders.

Steem is operating far from equilibrium but my following still apply. The whole point of steem is those stakeholders who have lots "skin in the game" get more say. Curation quality is not ignored, there are curation rewards for a reason.

"Curation mistakes" maybe to you but everyone is able to exercise their stake how they see fit. There is not really such thing as a curation mistake.

It was a great thought, but it doesn't seem to work on SteemIt. People are almost completely dependent on these heavy investors. Unfortunately HF19 isn't going to change much i'm afraid, but we'll see.

The fact that those large stakeholders have so much influence is problematic, they should be able to profit another way or additional algorithms have to be implemented to make it more fair. It's a hard problem, but i'm sure they're working on it.

Curation quality is almost completely ignored because the algorithms aren't functioning properly. Sometimes you can earn more curation rewards by voting on worse content than when you vote on fantastic content. Ideally it should be a self regulating system where curators could profit from fixing injustice instead of profiting from contributing to the injustice.

With curation mistake i essentially mean you voted differently than the average consensus. For example when you voted on a post that receives flags, you should get penalized slightly for voting on something that is rated as less than "perfect" content. Or when you end up being the only upvoter against 9 flags you should get penalized heavily (more than just losing out on curation rewards). The system needs an extra variable similar to reputation (but then actually significant haha).

Well there are curation mistakes when you would be talking about executing an optimal probabilistic strategy to maximize profits ;)

Well HF19 was sure a change.

Large stakeholders are going to be a part of any system. Nothing problematic directly.

Again, as far as steem is concerned - there is no curation mistakes. That is by design.

Sure hot and trending suck. But that has less to do with curation quality but the lack of quality content as a whole.

It should literally be the other way around, someone with 100x more money should only have 10 times more influence.

That can't work since people will split their stake. If 100x more money gives 10x more influence then you split it into 100 pieces and have 100x influence instead. The flattest we can go is linear. I agree that quadratic was nutty though

Ah good point, but my statement was just to make a point anyway ;)

BTW, are you stalking me now? haha

A great explanation. I have wavered back and forth on this position (still wavering, btw)

this is not the final implementation

the great thing about this beta process is how thorough it is, and the community's willingness to test out various cogs on the wheel

Haha yeah, you mean the various wheels and their various cogs.

Steem economics is extremely complex, really not sure how this will fair in the long run.
I would say that the system is operating far from steady state at the moment.

I don't think it will ever be "fair"

That's subjective to begin with, and we have folks with different motives to be here.

It will never really be fair unless I get the entire payout pool every day LOL

I have not followed the upcoming potential changes to the current voting system, I liken it to DC, why would they (politicians) ever vote for term limits to end their reign? So why would those making huge sums ever want to change the system?
When I first joined last month I did notice constant schemes to get noticed and the best system to "buy" upvotes, if that much energy was spent encouraging members to recruit new members, I believe that would be a better use of energy and resources, than paying out huge sums for whalecake recipes.
Once I commented on a high-value post (one of the good ones) and saw an awful lot of whale-ass kissing, like giddy children trying to get a parent to notice them....
If you look at a board, like the Theranos board of directors or any other big company, that top leadership clearly is looking out for their own self-interests.. This isn't a complaint, just an observation..

Well they have incentives to ensure the system benefits users or the users will leave. This is where a lot of the complexity resides.

There will be ass kissing and everything between.

I don't see ass kissing getting 10 dollars much anymore.

Theranos? You mean the diagnostics scam? Yeah sure the were looking out for their asses.

Everyone's cheering already but We still have to see what kind of impact this has. Hoping for the best.

Yeah idk yet. I sure like. You sure like it. But we don't really have much say.

Wow that's really interesting for me. Didn't now that there is such a complex system behind this :) Thanks for sharing and I am also looking forward what happen next..

Well linear rewards are nice for now :P

When is HF19 supposed to happen?

This is one of the 1st comments after HF19 but, speaking as someone at the bottom of the food chain who hasn't had any whale-support (not that I know of) payout results are certainly impressive.

What remains to be seen is how voting patterns will change now over the next few weeks.

With that said, there are some VERY clever people here and sooner or later, they'll find a way to "adjust" the odds in their favor.

I hope the assumptions that linear voting will make people consider their votes more carefully holds true.

Maybe you should recommend quadratic voting for public elections?

See exactly. People haven't adjusted their voting strategies. All those posts you see bumping up in value might not have had votes in the first place.

Time is the only thing that will tell.

Darwin at work?

I just joined Steem and noticed payouts don't depend on just votes; was planning to look up the algo used. Your post seems a good starting point. Nicely put!

Definitely try and read the white paper for information about how voting works. Right now your vote is directly proportional to your steem power.

I had no idea HF 19 was so well though. I had been looking for this information all over. Thanks for sharing!

very interesting post , thanks for sharing it !

Thanks for your effort to put this information here..great to know..upvoted...